Help Center> ModelArts> Tool Guide> Model Training> Submitting Training Jobs with Different Names
View PDF

Submitting Training Jobs with Different Names

In ModelArts, once a training job is submitted, the job name cannot be changed. Each time a training job is submitted, a new version of the training job is created. To submit training jobs with different names, perform the following operations:

Submitting a Training Job with a Different Name as the Existing Training Job

Assume that a training job named train-mnist has been submitted for handwritten digit recognition. When the job is submitted for the first time, the job version is V0001. After the job is executed, the algorithm code is optimized, and the job is submitted again, the system creates the V0002 version under the train-mnist job.

Figure 1 Submitting a training job

If you have another training algorithm, for example, an object detection algorithm, and want to submit a training job named dog-cat-recognize, perform the following steps:

  1. Click the plus sign in the upper right corner of the Edit Training Job Configuration dialog box to add a training job configuration. After setting the parameters, click Apply and Run to submit the training job to ModelArts.
    Figure 2 Adding a new training job configuration
  2. When you open the Edit Training Job Configuration page again, two job names are displayed in the Job Name drop-down list. You can click a job name to view the job configuration.
    Figure 3 Two job names in job configurations

Deleting a Training Job

If multiple training jobs with different names exist and you do not need to retain the local training job configuration, click the minus sign in the upper right corner to delete the local job configuration.

This operation will delete only the local configuration and will not delete jobs on the ModelArts management console.

Figure 4 Deleting a training job configuration

After you click the minus sign, a confirmation dialog box is displayed. After confirming that the information is correct, click Yes to delete the local job configuration.

Deleted job configurations cannot be restored. Exercise caution when deleting them.

Parent topic: Model Training